Annual Report 2022/Story: Consolidation of servers and email systems

Möjliggörande

2022

In this story, we focus on the work involved in cleaning up among our technical systems in order to reduce the maintenance of servers and e-mail systems. Our goal has been reducing long-term costs while ensuring a high level of operational reliability.

We have been cleaning our servers during the year.

We have a number of systems at Wikimedia Sverige that we use and maintain for both internal and external use. These include, for example, a start page for Wikipedia in Swedish, a map for Wiki Loves Earth and a tool to collect visitor statistics. In the past, we have also maintained several systems ourselves: our website and internal systems for work documents, mailing lists and time-reporting. Combined with the restructuring required by GDPR, we had a variety of systems and tools spread across a number of servers.

We haven't really had working routines for how those systems should be maintained; it was done as needed, which often meant when something was already broken (see the 2020 annual report). Nor have we had enough resources or competence to keep up with the systems. This has led to them not being safe, and incidents happened (see the 2019 annual report).

We have been working for some time on simplifying the work with our systems and the servers they are hosted on. The first step was to identify which systems we had and which of them we actually needed to keep. Then we investigated which supplier would be most suitable to have our new servers with. This choice was based on several parameters such as price, user-friendliness and environmental impact (see Memo on consolidation of cloud service providers).

Sooner or later, every server requires some loving care and attention. The more servers we have with different versions of software installed, the more maintenance is required. When you take into account that that every provider has their own system for settings, payments and more, work piles up. However, when you consolidate systems like we have done, you don't only save working time. When there are fewer different components to keep track of, safety improves as well.

We also changed the system that runs our internal mailing lists. In the past we had used Mailman, but after the old server that hosted it was show down, we chose to move the mailing lists to Google Workspace. Mailman was one of the few FOSS systems that we used and maintained ourselves as an association. Still, we felt that the practical benefits outweighed it, as now we could manage the mailing lists in a system we already used and where the users already had accounts (see Memo on migration of mailing lists).

Most of our systems have now been moved to our new servers and the last of the old servers will be shut down soon. This means that we now have a good starting point; what remains is to develop a routine to take care of our systems in their new forms. Once that's in place, it will make it easier for us to effectively maintain our systems and minimize the risk of maintenance being delayed, leading to unpleasant surprises in the future.